Light Availability and Distribution Patterns in a Tropical Premontane Wet Forest
Bryan
1Texas
1
Tarbox ,
Tomasz
2
Falkowski
A&M University (TAMU), 2State University of New York at Binghamton, 3University of Wisconsin at Madison, TAMU Soltis Center for Research and Education
Introduction
Light availability plays an important role in how plants
participate in the hydrologic and carbon cycles. One way
of looking at light availability is by leaf area index (LAI),
which is the measure of leaf area per unit land area. LAI
can be used to determine primary productivity, which
dictates carbon consumption. Further, LAI is important
for models which predict land surface-vegetation
feedbacks, because it can be used to parameterize
variables such as albedo and surface conductance.
Transpiration is a key hydrologic process which is
modified by factors such as LAI, because LAI
determines how much radiation filters through the
canopy. LAI is correlated with other structural features
such as DBH, height and stem basal area. These
features change with stand age and may vary
dramatically under different management regimes. In
addition to light availability, canopy structure also
dictates the flow of water, partitioning precipitation into
interception loss, throughfall and stemflow.
Research Goals

and Kevin
3
Davis
Characterize the relationships between structure and
light availability under different land uses.

Test the ability of hemispherical photography and
densiometers to accurately measure light availability.

Account for the throughfall portion of precipitation in a
plot where transpiration studies are simultaneously
occurring.
Results and Discussion
Mean tree height did not correlate well with either LAI or gap fraction
estimates. Mean DBH correlated well, but Stem Basal Area was both the
easiest characteristic to measure and the most accurate (Figure 1).
FPAR estimates were then used to generate light availability
maps for each site (Figure 5).
Figure 5. Light availability as
determined by the fraction of
photosynthetically active radiation
(PAR) transmittance penetrating
through the canopy. Peaks represent
areas of high light availability.
Throughfall precipitation varied between
gauges, due to the spatial heterogeneity
of the canopy (Figure 7). All of the gauges
measured similar patterns, however,
following net precipitation in a delayed
manner (Figure 8). On average,
throughfall represented 70% of net
precipitation for the month of July.
Figure 1. Stem basal area as estimated by wedge prism compared to leaf area
index (LAI) and gap fraction.
Estimates of gap fraction derived from spherical densiometer readings and
hemispherical photography were similar in all three plots. However, in the
primary forest there was a poor correlation between the two methods due to
operational error at the southwest point (Figure 2). Gap fraction as estimated
by densiometer and HemiView both correlated well with light availability as
measured by FPAR (Figure 3), though densiometer estimates were both
easier to obtain and more accurate. Leaf area index as estimated by
hemispherical photography matched the LAI value derived from FPAR
calculations very closely. However, within natural forest settings HemiView
dramatically underestimated LAI (Figure 4). This is probably due to leaf
clumping, which has been documented to induce underestimation of LAI.
Figure 7. Throughfall as a percentage of net precipitation
at the secondary forest plot.
Vertical light profiles showed a general trend of increasing
light availability with height, though there were numerous
discrepancies. This is due to gaps in the canopy, which
allow light to penetrate in at an angle, as well as structural
diversity (Figure 6). The ability to accurately portray these
profiles was restricted by the height of the pole, which only
extended to half the height of the larger trees in the plot.
Methods

Three plots were selected on or near the property of
the Soltis Center on the Caribbean slope of the
Cordillera Tilaran in Costa Rica. These plots were
representative of different land uses: a monoculture
tree plantation, a secondary forest and a primary
forest.

DBH, tree height and stem basal area were measured
with diameter tape, a laser clinometer and a 3 factor
wedge prism, respectively.

Hemispherical photographs and densiometer readings
were taken from 4 points within each plot.

PAR measurements were taken with a 1-m line
quantum sensor on a grid of transects for each plot
and compared to simultaneous PAR measurements
from a nearby clearing. LAI was estimated by using
the Beer-Lambert law (LAI = ln ((under canopy
PAR/reference PAR) / -κ)).


At the secondary forest plot, vertical light profiles were
measured at each of the 4 photograph points by
extending a quantum sensor on a 50-ft height pole.
Six tipping bucket rain gauges were placed randomly
throughout the secondary forest plot to measure
throughfall.
Figure 2. Comparison of methods
for estimating gap fraction.
Figure 3. Comparison of
gap fraction and FPAR
under three different land
use classes.
Figure 4. Comparison of LAI
under three different land use
classes. Hemispherical
photography underestimated LAI
in more natural settings.
Figure 8. Precipitation measured during a rain event from
six different tipping bucket gauges beneath the canopy in
the secondary forest plot, compared to net precipitation as
measured from outside the canopy.
Figure 6. Vertical light availability profiles for the secondary forest plot and
distribution of LAI.
Conclusions
Estimating stem basal area with a wedge prism appeared to be the easiest and most effective way to predict the light availability in any given stand. Using a spherical densiometer to estimate gap
fraction is a cheaper, easier and more accurate method than hemispherical photography, though it isn’t any more effective at obtaining LAI estimates. Differences in light availability and stem basal
area were more significant between natural and agricultural settings than between primary and secondary forest. DBH increased linearly with stand age.
Due to time constraints and involvement in concurrent projects at the site, not enough samples were taken to accurately portray each site, and some sites were altogether left out of measurements
such as vertical light profiles and throughfall. Additionally, stemflow was not measured, which is necessary to arrive at an estimate for interception losses. LIDAR scans were taken of each plot,
though the scans of the most otherwise studied plot (secondary forest) were not viable, due to registration errors. Coupling LIDAR models with this structural data should be a focus of future studies.
Acknowledgements
Research was supported by a grant from the National Science Foundation. Field and lab assistance was provided by Mark Tjoelker, Georgianne W. Moore, Robert Washington-Allen, Tony Cahill, Eugenio Gonzalez, Alberth Rojas and Chris Houser.